Remote Sensing Applications for Landslide Monitoring and Investigation in Western Canada

Review Remote Sensing Applications for Landslide Monitoring and Investigation in Western Canada

Renato Macciotta * and Michael T. Hendry

Department of Civil and Environmental Engineering, University of Alberta, Edmonton, AB T6G 2R3, Canada; [email protected] * Correspondence: [email protected]

Abstract: Transportation infrastructure in mountainous terrain and through river valleys is exposed to a variety of landslide phenomena. This is particularly the case for highway and railway corridors in Western Canada that connect towns and industries through prairie valleys and the Canadian cordillera. The fluidity of these corridors is important for the economy of the country and the safety of workers, and users of this infrastructure is paramount. Stabilization of all active slopes is financially challenging given the extensive area where landslides are a possibility, and monitoring and minimization of slope failure consequences becomes an attractive risk management strategy. In this regard, remote sensing techniques provide a means for enhancing the monitoring toolbox of the geotechnical engineer. This includes an improved identification of active landslides in large areas, robust complement to in-place instrumentation for enhanced landslide investigation, and an improved definition of landslide extents and deformation mechanisms. This paper builds upon the extensive literature on the application of remote sensing techniques and discusses practical insights gained from a suite of case studies from the authors’ experience in Western Canada. The review of the case studies presents a variety of landslide mechanisms and remote sensing technologies. The aim of the paper is to transfer some of the insights gained through these case studies to the reader. Keywords: remote sensing; landslide monitoring; geotechnical investigation; risk management Citation: Macciotta, R.; Hendry, M.T. Remote Sensing Applications for Landslide Monitoring and Investigation in Western Canada. 1. Introduction Remote Sens. 2021, 13, 366. https://doi.org/10.3390/rs13030366 The integrity of the transportation network in Western Canada for securing uninter- rupted fluidity of goods and people through the network are essential for the economy of Received: 16 December 2020 the region. Transportation corridors through the Canadian Cordillera and through river Accepted: 19 January 2021 valleys in the prairies are exposed to natural slope instabilities and steep cuts required to ac- Published: 21 January 2021 commodate railroads and highways, where slope-related hazards are common. Landslide types in these regions include rock falls, rock slides and topples, earth flows, deep seated Publisher’s Note: MDPI stays neutral rotational and translational soil landslides, and compound landslides [1–6]. In this regard, with regard to jurisdictional claims in a robust approach to manage the risks associated with landslides in natural and cut slopes published maps and institutional affil- becomes essential. iations. The risk management framework for landslides in natural and cut slopes has been previously discussed elsewhere [7–11]. After defining the scope of analysis, the framework requires characterizing the magnitude and likelihood of the slope failure, as well as the magnitude and likelihood of the potential consequences, which become the basis for esti- Copyright: © 2021 by the authors. mating risk (combination of likelihood and consequence for each failure scenario analyzed). Licensee MDPI, Basel, Switzerland. The level of risk calculated in this risk analysis (qualitative or quantitative) is evaluated This article is an open access article against some adopted criteria to define its tolerability (risk assessment). The outcome of distributed under the terms and the risk assessment informs decision-making regarding the requirement for risk mitigation conditions of the Creative Commons and control strategies (e.g., protection, stabilization, early warning), as well as the need Attribution (CC BY) license (https:// for increased knowledge of the slope physical characteristics (e.g., geology, groundwater), creativecommons.org/licenses/by/ failure mechanisms and potential triggers; and monitoring requirements for managing 4.0/).

Remote Sens. 2021, 13, 366. https://doi.org/10.3390/rs13030366 https://www.mdpi.com/journal/remotesensing Remote Sens. 2021, 13, x FOR PEER REVIEW 2 of 22 Remote Sens. 2021, 13, 366 2 of 23 warning), as well as the need for increased knowledge of the slope physical characteristics (e.g., geology, groundwater), failure mechanisms and potential triggers; and monitoring requirements for managing residual risks. This is where remote sensing techniques residual risks. This is where remote sensing techniques become provide increased beneﬁt to become provide increased benefit to in-place instrumentation, particularly given the more in-place instrumentation, particularly given the more frequent adoption of quantitative risk frequent adoption of quantitative risk analyses since the late 1990s and early 2000’s [12– analyses since the late 1990s and early 2000’s [12–15]. This risk management framework is 15]. This risk management framework is illustrated in Figure 1 [11]. illustrated in Figure1[11].

FigureFigure 1. 1.General General risk risk management management framework framework for for landslides landslides in in natural natural and and cut cut slopes. slopes. After After [ 11[11].].

WithinWithin the the risk risk management management framework framework in in Figure Figure1, 1, monitoring monitoring plays plays a fundamentala fundamental role.role. NotNot onlyonly is is robust robust monitoring monitoring the the cornerstone cornerstone of of effective effective early early warning, warning, but but also also fundamentalfundamental for for increased increased understanding understanding of of the the landslide landslide mechanisms, mechanisms, kinematics kinematics and and triggers.triggers. InIn thisthis regard,regard, remoteremote sensingsensing technologies technologies havehave continued continued to to develop develop to to the the pointpoint where where they they are are becoming becoming routine routine practice practice for for landslide landslide monitoring monitoring and and for for landslide landslide investigation as part of the hazard analysis in Figure1[ 16]. These include ground-based investigation as part of the hazard analysis in Figure 1 [16]. These include ground-based and airborne (from manned and unmanned aircrafts) photogrammetry, laser scanning, and airborne (from manned and unmanned aircrafts) photogrammetry, laser scanning, and satellite and ground-based radar interferometry [17–28]. and satellite and ground-based radar interferometry [17–28]. This paper presents a review of six case studies in Western Canada from the au- This paper presents a review of six case studies in Western Canada from the authors’ thors’ experience; where remote sensing was used in combination with other site inves- experience; where remote sensing was used in combination with other site investigation tigation and monitoring technologies to gain understanding of the slope deformation and monitoring technologies to gain understanding of the slope deformation mechanisms mechanisms and develop early warning systems. A brief description of the geologic and develop early warning systems. A brief description of the geologic context, landslide context, landslide extents and deformation patterns are followed by a description of the extents and deformation patterns are followed by a description of the instrumentation and instrumentation and remote sensing techniques deployed and the insights gained. It is remote sensing techniques deployed and the insights gained. It is noted that the work noted that the work presented here reflects the methods that were considered most suited forpresented the case here studies reflects discussed, the methods which that may were vary considered in other contexts most suited and for the other case jurisdic- studies tions.discussed, This workwhich is may not anvary exhaustive in other contexts literature and review for other of remote jurisdictions. sensing This techniques work is for not landslidean exhaustive investigation; literature however, review of it remote presents sensing valuable techniques examples for for landslide how the investigation; information capturedhowever, from it presents remote sensingvaluable techniques examples enhance for how the the implementation information captured of the landslide from remote risk managementsensing techniques framework. enhance the implementation of the landslide risk management framework. 2. Chin Coulee Landslide 2.1.2. Chin Location Coulee and GeologyLandslide 2.1. TheLocation Chin and Coulee Geology landslide is located near the town of Taber in southern Alberta, Canada.The TheChin landslide Coulee islandslide within ais broadlocated valley near thatthe wastown dammed of Taber to in develop southern the Alberta, Chin CouleeCanada. water The reservoir.landslide Theis within landslide a broad was likelyvalley triggered that was by dammed a combination to develop of the the works Chin requiredCoulee water to relocate reservoir. Highway The landslide 36 from was near likely the toe triggered of the slope by a combination to the crest ofof thethe slopeworks (whichrequired required to relocate fill placementHighway 36 in from the upper near the part toe of of the the slope) slope and to the filling crest of of the the reser- slope voir(which[29– required31]. The locationfill placement of the in Chin the Couleeupper part landside of the and slope) an and aerial filling view of are the shown reservoir in Figure2 (modified from [30]).

Remote Sens. 2021, 13, x FOR PEER REVIEW 3 of 22

[29–31]. The location of the Chin Coulee landside and an aerial view are shown in Figure 2 (modified from [30]). The local stratigraphy of the landslide is shown in Figure 2. Medium plasticity clay fill material from the initial highway construction and subsequent realignments and regrading is exposed at the head scarp of the landslide. A silty clay till with traces of fine gravel, mantles the valley slope. The till has low to medium plasticity and is very stiff to hard. The thickness of the till varies from approximately 20 m at the toe of the slope, to 35 m near the head scarp. Highly fractured shales and coal were encountered underneath the till [31]. The interpreted piezometric elevation is also shown in Figure 2. The landslide stratigraphy and piezometric elevation have been interpreted based on borehole logs (BH- Remote Sens. 2021, 13, 366 1 to BH-7 in Figure 2), inclinometer readings (in BH-2, 3, 4 and 5), and piezometric3 of 23 readings (BH-1, 6 and 7) [30].

FigureFigure 2.2. Location,Location, aerialaerial view view ( a(a)) and and typical typical geologic geologic cross cross section section (b ()b of) of the the Chin Chin Coulee Coulee landslide. Adaptedlandslide. from Adapted [30]. from [30].

2.2. LandslideThe local Extents stratigraphy and Deformation of the landslide Patterns is shown in Figure2. Medium plasticity clay ﬁll material from the initial highway construction and subsequent realignments and regrading The landslide extents were interpreted based on field and photographic mapping of is exposed at the head scarp of the landslide. A silty clay till with traces of ﬁne gravel, the surface expression of the lateral failure surfaces and the head scarp. The landslide is mantles the valley slope. The till has low to medium plasticity and is very stiff to hard. The thickness of the till varies from approximately 20 m at the toe of the slope, to 35 m near the head scarp. Highly fractured shales and coal were encountered underneath the till [31]. The interpreted piezometric elevation is also shown in Figure2. The landslide stratigraphy and piezometric elevation have been interpreted based on borehole logs (BH-1 to BH-7 in Figure2), inclinometer readings (in BH-2, 3, 4 and 5), and piezometric readings (BH-1, 6 and 7) [30].

2.2. Landslide Extents and Deformation Patterns The landslide extents were interpreted based on ﬁeld and photographic mapping of the surface expression of the lateral failure surfaces and the head scarp. The landslide is approximately 350 m wide and up to 45 m deep. The length of the landslide is approximately Remote Sens. 2021, 13, 366 4 of 23

200 m long with the toe of landslide within the reservoir. The failure surface has been inferred based on the location of the head scarp, the shear zones identiﬁed in the slope inclinometers, and the location of a low-strength coal seam within the shales and siltstones (Figure2). The total landslide volume is estimated at approximately 2 million cubic meters, based on measurement along the proposed failure plane and current landslide extents. The Chin Coulee landslide has been interpreted as a translational retrogressive landslide seated within the highly fractured bedrock zone and sliding over the low-strength coal seam in Figure2. Retrogression is inferred from surface tension cracks and concentrated deformations measured in the slope inclinometers near the head scarp. Displacement rates vary between 10 and 50 mm/year.

2.3. Instrumentation and Remote Sensing Techniques Inclinometer readings are available for BH-2, 3, 4 and 5. Other instrumentation at the site includes a differential GPS system consisting of ten units (installed in July of 2018). This system is acquiring high frequency data, and details of its installation and interpretation can be found in [29]. Here, we will focus on the use of remote sensing techniques to obtain quantitative landslide information by leveraging historical aerial photography and state of practice remote sensing acquisition and processing. Historic stereo pairs were available for 1945, 1960, 1970, 1982, 1993, 1999 and 2012. The stereopairs in 1982 had the more detailed scale (1:2500), with five photographs over- lapping the scene of the landslide. These photographs were used to develop a surface model of the landslide using the software Pix4D (www.pix4d.com), and validated against the location of large boulders outside the landslide area that were assumed to be stable. The surface model was converted (sampled) to a point cloud with a minimum density of 500 points/m2. Laser scanning of the landslide was initiated in July of 2018 using a long-range scanner, ILRIS-LR, with a wavelength of 1064 nm, a pulse frequency of 10 kHz, and a beam divergence of 0.014324◦ [32]. Scanning was done from 3 locations across the reservoir from the landslide [29], with the average scan distance on Chin Coulee at 1000 m. The system has an accuracy of ±7 mm, which can be improved to ±4 mm through shot repetition and an averaging option. The scanner rendered over 8 million points, after removing tall vegetation and artificial structures and equipment. Point density ranged between 1000 and 2000 points/m2. Quantitative information of the amount of slope deformation since the movement initiated was sought as a means to improve the understanding on the landslide kinematics for a better interpretation of the mechanisms behind the slope movement. Changes in the slope surface were calculated through comparing the point clouds from the 1982 aerial photographs and a laser scan from July 2019 (Change Detection). The software used for point cloud analysis and change detection was CloudCompare V2.10 (available through www.cloudcompare.org). The method used for change detection was M3C2 [33]. In the M3C2 method, change is quantified as the distance between point clouds measured parallel to the normal vector of the surface in the vicinity of each measurement point. The threshold for change detectability (limit of detection—LOD) was set at two standard deviations of measurements within areas considered stable (zero change). Changes calculated within ± two standard deviations were considered potential measurement randomness. The limit of detection between the 1982 photogrammetric model and the 2019 LiDAR model was 1 m, attributed to a lower accuracy of the 1982 model derived from historic photos and surficial changes between 1982 and 2019 due to erosional processes [30].

2.4. Results and Discussion The results of the Change Detection between the 1982 photographs and the laser scanning in 2019 are shown in Figure3[ 30]. Negative values correspond to material loss and positive values correspond to movement towards the scanner (out of slope). Remote Sens. 2021, 13, x FOR PEER REVIEW 5 of 22

The results of the Change Detection between the 1982 photographs and the laser scanning in 2019 are shown in Figure 3 [30]. Negative values correspond to material loss and positive values correspond to movement towards the scanner (out of slope). Loss of material in the upper half of the slope was measured at approximately 4 m between 1982 and 2019, which is validated by inspection of the landslide scarp when compared to a 2.5-m-high fence and a 1.5-m-long guardrail post (Figure 3a). Although there are differences in resolution and accuracy between the models, this comparison increased the confidence in the magnitudes of the calculated change for interpretation of the landslide kinematics. Changes in the lower half of the landslide suggest 2 to 3 m of outward movement at the toe of the slope (and up to 4 m in some areas). Unlike the gradational change in the direction of deformation expected from rotational landslides, the transition is sharp between the areas showing material loss (interpreted as downward movement of the landslide mass) and the outward displacement in the lower half of the landslide. This deformation trend is consistent with compound landslides mechanisms, where an active wedge moves downward (seen in the change as a loss of material) pushing a passive wedge outward on a sub-horizontal shear zone (seen in the calculated change as gain in material or movement towards the scanner). The sharp change to outward movement suggests the location of the contact between the driving and passive wedges (Figure 3) [30]. This enhanced knowledge regarding the mechanisms of the Chin Coulee Landslide Remote Sens. 2021, 13, 366 allows improved determination of the slope stability as the reservoir fluctuates and5 of 23 evaluate the potential for sudden acceleration under reservoir increases and draw down conditions, supporting landslide risk evaluation under diverse scenarios.

FigureFigure 3. Change 3. Change detection detection results at results the Chin at the Coulee Chin landslide Coulee landslide(a) and interpreted (a) and interpreted landslide landslide mechanismmechanism (b). Adapted (b). Adapted from from[30]. [30].

3. Site C018Loss of material in the upper half of the slope was measured at approximately 4 m 3.1. Locationbetween and 1982 Geology and 2019, which is validated by inspection of the landslide scarp when compared to a 2.5-m-high fence and a 1.5-m-long guardrail post (Figure3a). Although there are differences in resolution and accuracy between the models, this comparison increased the conﬁdence in the magnitudes of the calculated change for interpretation of the landslide kinematics. Changes in the lower half of the landslide suggest 2 to 3 m of outward movement at the toe of the slope (and up to 4 m in some areas). Unlike the gradational change in the direction of deformation expected from rotational landslides, the transition is sharp between the areas showing material loss (interpreted as downward movement of the landslide mass) and the outward displacement in the lower half of the landslide. This deformation trend is consistent with compound landslides mechanisms, where an active wedge moves downward (seen in the change as a loss of material) pushing a passive wedge outward on a sub-horizontal shear zone (seen in the calculated change as gain in material or movement towards the scanner). The sharp change to outward movement suggests the location of the contact between the driving and passive wedges (Figure3)[ 30]. This enhanced knowledge regarding the mechanisms of the Chin Coulee Landslide allows improved determination of the slope stability as the reservoir ﬂuctuates and evaluate the potential for sudden acceleration under reservoir increases and draw down conditions, supporting landslide risk evaluation under diverse scenarios.

3. Site C018 3.1. Location and Geology Site C018 is a weathered rock slope adjacent to Highway 837 along the south bank of the Red Deer River, approximately 10 km northwest of the town of Drumheller, Alberta (Figure4) . Remote Sens. 2021, 13, x FOR PEER REVIEW 6 of 22

Site C018 is a weathered rock slope adjacent to Highway 837 along the south bank of the Red Deer River, approximately 10 km northwest of the town of Drumheller, Alberta (Figure 4). This slope is predominantly in a weak, poorly indurated sedimentary rock with layers of stronger sandstone, siltstone, and coal [34]. Bedrock is weak to extremely weak, with a UCS of less than 5 MPa and GSI of 20 to 25. The exposed bedding planes are visible in the valley slopes due to the variations of color and texture of the different materials. Bedding structures have an approximately sub-horizontal inclination of 10° [35]. The slope at site C018 shows extensive signs of weathering processes including deep gullies, sinkholes, weak runoff marks, and pipe erosion, where weathering processes are ubiquitous and result in a soft soil-like material at surface, with the bedrock structure almost eradicated. Discontinuous blocks of more indurated bedrock remain as less Remote Sens. 2021, 13, 366 weathered rock blocks [34]. 6 of 23

3.2. Landslide Extents and Deformation Patterns Site C018 is a slope approximately 520 m long and 90 m high (Figure 4). This site has This slope is predominantly in a weak, poorly indurated sedimentary rock with layers of a long history of instabilities with fallen material reaching and blocking the highway. The stronger sandstone, siltstone, and coal [34]. Bedrock is weak to extremely weak, with a slope displays three modes of failure, which largely depend on the moisture conditions of UCS of less than 5 MPa and GSI of 20 to 25. The exposed bedding planes are visible the materials forming the slope: (1) earthflows associated with heavy rainfall; (2) rock falls in the valley slopes due to the variations of color and texture of the different materials. that result from differential weathering rates and undermining◦ of various rock units; and Bedding structures(3) slides have anof frozen approximately slabs of highly sub-horizontal weathered inclination rock [34]. of 10 [35].

Figure 4. LocationFigure of the 4. study Location site inof Canadathe study (lower site in right) Canada and (low aerialer right) view and of Site aerial C018. view of Site C018.

The slope at3.3. site Remote C018 showsSensing extensiveTechnique signsand Analyses of weathering processes including deep gullies, sinkholes, weakSite runoffC018 presents marks, andan example pipe erosion, of the where use of weathering inexpensive processes unmanned are aerial vehicle ubiquitous and result(UAV) in photogrammetry a soft soil-like material as the atbasis surface, for monitoring with the bedrock surface structure changes almostand for developing eradicated. Discontinuousvirtual surface blocks models of more for geomorphic indurated bedrockanalysis remain(in this asexample, less weathered preferential drainage rock blocks [34].paths and rock fall trajectory models). Three UAV photogrammetry surveys were conducted in December 2017, May 2018 3.2. Landslide Extentsand November and Deformation 2018, Patternsrespectively. The UAVs used in the surveys had a 12-MP (mega Site C018 ispixels) a slope camera approximately for the first 520 and m third long flight and and 90 ma 17-MP high (Figure camera4 for). Thisthe second site flight. The has a long history12-MP of instabilities camera had with a sensor fallen 1/2.3 material″ CMOS reaching (6.3 mm and width blocking and 4.7 the mm highway. height), a 94° field The slope displaysof threeview, modesa 20 mm of focal failure, length, which and largely aperture depend f 1:2.8. on The the 17-MP moisture camera conditions had a sensor 1/2.3″ of the materialsCMOS forming (6.3 the mm slope: width (1) and earthﬂows 4.7 mm height), associated an 82° with field heavy of view, rainfall; a 44 mm (2) rockfocal length, and falls that result fromaperture differential f 1:3.3. Minimum weathering overlap rates andbetween undermining photographs of various was 60%, rock which units; required 655, and (3) slides of frozen slabs of highly weathered rock [34].

3.3. Remote Sensing Technique and Analyses

Site C018 presents an example of the use of inexpensive unmanned aerial vehicle (UAV) photogrammetry as the basis for monitoring surface changes and for developing virtual surface models for geomorphic analysis (in this example, preferential drainage paths and rock fall trajectory models). Three UAV photogrammetry surveys were conducted in December 2017, May 2018 and November 2018, respectively. The UAVs used in the surveys had a 12-MP (mega pixels) camera for the first and third flight and a 17-MP camera for the second flight. The 12-MP camera had a sensor 1/2.300 CMOS (6.3 mm width and 4.7 mm height), a 94◦ field of view, a 20 mm focal length, and aperture f 1:2.8. The 17-MP camera had a sensor 1/2.300 CMOS (6.3 mm width and 4.7 mm height), an 82◦ field of view, a 44 mm focal length, and aperture f 1:3.3. Minimum overlap between photographs was 60%, which required 655, 600 and 910 photographs for each of the three surveys in chronological order. Surface models Remote Sens. 2021, 13, 366 7 of 23

Remote Sens. 2021, 13, x FOR PEER REVIEW 7 of 22

600 and 910 photographs for each of the three surveys in chronological order. Surface were developed using the software Pix4D and the surface model was converted (sampled) models were developed using the software Pix4D and the surface model was converted to a point cloud with a minimum density of 500 points/m2 for Change Detection using (sampled) to a point cloud with a minimum density of 500 points/m2 for Change Detection the M3C2 methodusing [35 the]. M3C2 The surfaces method [35]. were The also surfaces used towere develop also used a surface to develop drainage a surface net- drainage work followingnetwork the methods following outlined the methods in [36]; andoutlined for 2.5-dimensional in [36]; and for rock 2.5-dimensional fall trajectory rock fall simulations followingtrajectory the simulations methods outlinedfollowing inthe [ 37methods] and [ 38outlined]. in [37] and [38].

3.4. Results and3.4. Discussion Results and Discussion The results obtainedThe results by [35 obtained] are summarized by [35] are in summariz Figure5. Thised in ﬁgure Figure shows 5. This the figure Change shows the Detection betweenChange December Detection 2017 between and November December 2018, 2017 theand surface November drainage 2018, modelthe surface on the drainage November 2018model surface on model,the November and a plan2018 viewsurface of model, rock fall and trajectories a plan view on of therock November fall trajectories on 2018 surface model.the November 2018 surface model.

Figure 5.Figure Change 5. DetectionChange Detectionbetween December between 2017 December and November 2017 and 2018 November (a), surface 2018 drainage (a), surface model drainage on the November model 2018 surface model (b), and plan view of rock fall trajectories on November 2018 surface model (c). After [35]. on the November 2018 surface model (b), and plan view of rock fall trajectories on November 2018 surface model (c). AfterThe [35 Change]. Detection results highlighted the state of activity of the areas of the slope that had recorded instabilities (Figure 5a–c). This analysis provided insight to prioritize The Changearea Detection F1 for further results hazard highlighted analysis, the as some state material of activity was of actively the areas being of eroded the slope mid-slope that had recordedand instabilitiesaccumulating (Figure near the5a–c). toe. This active analysis erosion provided was more insight pronounced to prioritize on weather- area F1 for further hazard analysis, as some material was actively being eroded mid-slope and accumulating near the toe. This active erosion was more pronounced on weather- susceptible rock units, undermining stronger rock above, which could lead to rock fall events with the potential to reach the highway. This information, in combination with the Remote Sens. 2021, 13, x FOR PEER REVIEW 8 of 22 Remote Sens. 2021, 13, 366 8 of 23 susceptible rock units, undermining stronger rock above, which could lead to rock fall events with the potential to reach the highway. This information, in combination with the rockrock fall fall trajectory trajectory simulations simulations in in Figure Figure5c, 5c allowed, allowed the the development development of of hazard hazard mitigation mitigation strategiesstrategies at at this this site. site. The The surface surface drainage drainage models models provided provided valuable valuable information information to evalu- to ateevaluate the impact the impact of surface of surface runoff onrunoff the potentialon the potential for earthﬂow for earthflow occurrences. occurrences. These results These suggestresults surfacesuggest watersurface management water management strategies strategies as one slope as one hazard slope and hazard risk and control risk option, control divertingoption, diverting surface runoff surface away runoff from away the from crest ofthe the crest slope. of the slope.

4.4. 10-Mile 10-Mile Slide Slide 4.1.4.1. Location Location and and Geology Geology TheThe 10-Mile 10-Mile Slide Slide is is an an active active portion portion of of a a post-glacial post-glacial earthﬂow, earthflow, and and is is located located in in the the ProvinceProvince of of British British Columbia, Columbia, Canada, Canada, North North of of Vancouver Vancouver [39 [39]] (Figure (Figure6a). 6a). The The landslide landslide hashas been been monitored monitored for for over over 40 40 years, years, as as its its deformation deformation has has led led to to frequent frequent maintenance maintenance ofof a a highway highway and and railway railway track track that that cross cross the the continuously continuously deforming deforming mass. mass.

FigureFigure 6. 6.Location Location and and front front view view of of the the 10-Mile 10-Mile Slide Slide (a ().a). Sketch Sketch of of typical typical section section and and stratigraphy ofstratigraphy the landslide of( bthe) After landslide [39]. (b) After [39].

TheThe landslide landslide is is within within the the Fraser Fraser River Rive valley,r valley, where where glacial glacial and post-glacial and post-glacial sedi- mentssediments are common are common along along the valleythe valley slopes. slopes. The The bedrock bedrock lithology lithology in in the the area area includes includes andesiteandesite to to dacite dacite volcanic volcanic rocks, rocks, sandstone, sandstone, and and shale. shale. These These rocks rocks are overlain are overlain by qua- by ternaryquaternary deposits deposits (including (including glacial glacial drift drift blankets, blankets, colluvium, colluvium, alluvium, alluvium, and and landslide landslide deposits) [39,40]. This is reﬂected in the stratigraphy of the landslide, shown in Figure6b,

Remote Sens. 2021, 13, 366 9 of 23

which consists of a layer (up to 20 m thick) of mixed landslide deposits including medium to high plastic clays, silts, and the presence of zones of varying sand and gravel content. This layer overlies colluvium materials (sand, silt, and clay), which in turn overlie glacial deposits (stiff, gravelly clayey, and silty till with the presence of glacioﬂuvial well graded, rounded gravel, and sand deposits).

4.2. Landslide Extents and Deformation Patterns The 10-Mile Slide is approximately 200 m wide and 140 m high, with an approximate volume of 750,000 m3 [40]. Since the 1980s, displacement rates have varied spatially and temporally, with recorded rates up to 10 mm/day. The area of deformation was inferred from sharp scarps and tension cracks. The railway operator installed a 128-m-long retaining wall to control deformations, to an approximate depth of 16.5 m. and the highway operator has frequently realigned the highway as a response to the landslide movement. The landslide is sliding on a continuous shear surface that extends upslope beyond the location of the railway tracks (Figure6). This continuous shear surface appears to have a dip of approximately 22◦, sub-parallel to the ground surface. Piezometers installed at different locations within the 10-Mile Slide have measured pore water pressure equal to or less than atmospheric pressure, suggesting that increases in pore pressures are not the driving mechanism for slope movements [39,40].

4.3. Instrumentation and Remote Sensing Techniques Several slope inclinometers (SI) have been installed at the site. These were typically sheared within weeks of installation due to the fast deformation velocities [39]. Survey prisms (19 in total) were installed along the retaining wall and have been monitored since 2011. These were manually surveyed at different frequencies, with daily measurements in periods when the landslide showed the fastest deformation velocities. A differential GPS system consisting of 11 GPS units was installed in April of 2017 (same system to the one installed at the Chin Coulee Landslide). This system acquires high frequency data, and details of its installation and interpretation can be found in [41]. Ground-based (also terrestrial) laser scanning (TLS) was conducted in May and July of 2017 using a long-range scanner, ILRIS-LR, the same equipment used at Chin Coulee. Scanning was done from two locations across the river, with the average scan distance approximately 600m and a point density of approximately 1000 points/m2 [41]. The point clouds from the TLS surveys were used for Change Detection with the software Cloud- Compare V2 and the M3C2 method (same procedures discussed before). The threshold for change detectability (limit of detection—LOD) was set at two standard deviations of measurements within areas considered stable; 5 cm in the case of the two scans.

4.4. Results and Discussion The results of the Change Detection at the 10-Mile Slide between May and July of 2017 are shown in Figure7. These results are shown together with the deformations measured by the GPS system between April and November of 2017. Figure7a shows the location of the GPS units, together with the deformation vectors (to a scale of 1:500) and Figure7b shows the cumulative deformations in selected units for the same period of time. GPS monitoring showed that left and central areas of the landslide (North West side) have the highest displacement magnitudes (GPS units 46, 47 and 48). At the center of the landslide and near the highway cut, Unit 46 showed 400 mm displacement between April and November 2017. This is an order of magnitude larger than the units at the south-west end of the landslide. Similarly, Unit 43 showed displacement of approximately 30 mm during the same time frame. Remote Sens. 2021, 13, x FOR PEER REVIEW 10 of 22

displacement was 75 mm, with a range of 30 to 90 mm with Change Detection. At Unit 47, the displacement was 80 mm, with a range of 60 to 100 mm using Change Detection. At Remote Sens. 2021, 13, 366 Unit 46, the cumulative displacement was 180 mm. However, Change Detection ranged10 of 23 between 50 and 125 mm. This difference near Unit 46 could be associated with the lower density of TLS points on this area due to the oblique angle of the LiDAR shot.

FigureFigure 7.7. Location ofof GPSGPS units andand displacement vectorsvectors atat aa scalescale ofof 1:500 (aa),), cumulativecumulative displacementsdisplacements ofof selectedselected unitsunits ((bb),), andand TLSTLS ChangeChange DetectionDetection betweenbetween May May and and July July 2017 2017 ( c(c)) After After [ 41[41].].

TheImportantly, results from theChange areas showing Detection increas are compatibleed deformation in magnitude within with the thelandslide GPS units are (consideringconsistent between the two-month both monitoring period between techniqu scans).es. TLS At Change Unit 48, Detection the cumulative allows for displace- better mentdefinition was of 75 the mm, extent with of a the range most of active 30 to areas 90 mm of withthe landslide, Change Detection.which allows At for Unit better 47, theinterpretation displacement of the was landslide 80 mm, kinematics. with a range It has of 60been to 100interpreted mm using that Change an active Detection. block at Atthe Unit left side 46, the of the cumulative landslide displacement toe is moving was out 180 of slope mm. However,at a faster rate, Change leading Detection to a decrease ranged betweenin the stability 50 and of 125 the mm. slope This mass difference behind, near which Unit tends 46 could to follow be associated the movements with the lowerof the densitylower, left of TLS block. points TLS on Change this area Detection due to the complemented oblique angle of with the LiDAR the detailed shot. temporal resolutionImportantly, of the GPS the areasUnits showing allows for increased better interpretation deformation of within the landslide the landslide mechanisms, are con- sistentand therefore between improved both monitoring modeling techniques. of potential TLS stabilization Change Detection strategies. allows The setup for better also deﬁnitionallows for ofthe the near extent real-time of the monitoring most active frequency areas of the required landslide, for landslide which allows early for warning better interpretationsystems. of the landslide kinematics. It has been interpreted that an active block at the left side of the landslide toe is moving out of slope at a faster rate, leading to a decrease in the5. Ashcroft stability Subdivision of the slope massand the behind, Ripley which Landslide tends to follow the movements of the lower, left5.1. block.Location TLS and Change Geology Detection complemented with the detailed temporal resolution of the GPS Units allows for better interpretation of the landslide mechanisms, and therefore Fourteen landslides have been identified along a 10 km stretch of the Thomson River improved modeling of potential stabilization strategies. The setup also allows for the near Valley, south of Ashcroft, British Columbia, Canada [42]. The location of this section of real-time monitoring frequency required for landslide early warning systems. valley is shown in Figure 8. 5. AshcroftEight soil Subdivision units overlie and andesitic the Ripley bedrock Landslide in the study area [43]. These units were 5.1.deposited Location by and the Geology advance and retreat of three glaciations (Figure 8). Unit 2 is comprised of highly plastic clay interbedded with glaciolacustrine silt (unit 2), heavily consolidated Fourteen landslides have been identiﬁed along a 10 km stretch of the Thomson River by glacial action. Unit 3 presents large quantities of silt, sand, gravel, and boulders near Valley, south of Ashcroft, British Columbia, Canada [42]. The location of this section of valleyits base, is shownalso consolidated in Figure8. by glacial action. Unit 6 corresponds to glacial till (gravel size particles in a sandy, silty and clayey matrix). The stratigraphy of the valley is further complicated by an ancient Pleistocene river, which had cut down into unit 2, to the east of

Remote Sens. 2021, 13, x FOR PEER REVIEW 11 of 22 Remote Sens. 2021, 13, 366 11 of 23 the current location of the Thompson River and deposits of boulders, gravels, and sands at depths below the interface of units 2 and 3 have been reported in borehole logs [42].

FigureFigure 8. 8.Location Location of of landslides landslides south south of of Ashcroft Ashcroft (a (),a), detailed detailed landslide landslide layout layout (b (),b), and and stratigraphy instratigraphy the area (c). in Adapted the area from(c). Adapted [42]. from [42].

5.2. LandslideEight soil Extents units overlieand Deformation andesitic Patterns bedrock in the study area [43]. These units were depositedThe landslides by the advance in the and10 km retreat section of three of the glaciations valley range (Figure in volume8). Unit between 2 is comprised 1 and 15 ofmillion highly m plastic3 and their clay velocities interbedded range with from glaciolacustrine 4 to 30 mm/year silt to (unit over 2), 100 heavily mm/year, consolidated although bysome glacial have action. had faster Unit 3displacement presents large episodes quantities of meters of silt, sand,per day gravel, to meters andboulders per hour near [42]. itsThe base, areal also extent consolidated of these landslides by glacial is action. shown Unit in Figure 6 corresponds 8b. These to slides glacial typically till (gravel develop size particles in a sandy, silty and clayey matrix). The stratigraphy of the valley is further as compound slides, with an initial sliding mass that translates sub horizontally along the complicated by an ancient Pleistocene river, which had cut down into unit 2, to the east of surfaces of rupture. The surfaces of rupture of the landslides through this section of the the current location of the Thompson River and deposits of boulders, gravels, and sands at Thompson River Valley appear to occur along a common weak layer of clay located below depths below the interface of units 2 and 3 have been reported in borehole logs [42]. the bottom of the river. After some movements, displaced masses become unstable and 5.2.active Landslide wedges Extents form andat the Deformation main scarps. Patterns The weight of these wedges drives the mass towards the river [42]. The landslides in the 10 km section of the valley range in volume between 1 and One of these landslides, the Ripley Landslide, has been extensively investigated and 15 million m3 and their velocities range from 4 to 30 mm/year to over 100 mm/year, shows an annual cycle of deformation accelerating in the Fall (September) and although some have had faster displacement episodes of meters per day to meters per decelerating in the Spring (April), as a response to changes in river elevation. This hour [42]. The areal extent of these landslides is shown in Figure8b. These slides typically landslide has shown annual deformations over 100 mm/year [42,44]. develop as compound slides, with an initial sliding mass that translates sub horizontally along the surfaces of rupture. The surfaces of rupture of the landslides through this section 5.3. Instrumentation and Remote Sensing Techniques of the Thompson River Valley appear to occur along a common weak layer of clay located belowOnly the bottoma subset of the of river.these After landslides some movements, have been displaced instrumented, masses most become with unstable slope andinclinometers active wedges that formare atmanually the main read, scarps. annu Theally. weight The ofRipley these wedgeslandslide, drives however, the mass has towardsbecome thea field river laboratory [42]. for investigation and monitoring. Monitoring instrumentation at thisOne landslide of these landslides,include grouted-in the Ripley vibratin Landslide,g wire has piezometers, been extensively inclinometer investigated casings, and showsShape anAccel annual Arrays cycle (SAA), of deformation Differential accelerating GPS systems, in the and Fall a (September)wave-guide andsystem decelerating that uses inwaves the Spring emitted (April), as the as landslide a response deforms to changes and in in riverterprets elevation. them Thisas relative landslide displacement; has shown annual deformations over 100 mm/year [42,44].

Remote Sens. 2021, 13, x FOR PEER REVIEW 12 of 22 Remote Sens. 2021, 13, 366 12 of 23 among others. Site investigation techniques deployed at the site include borehole logging, geophysical investigations (seismic, electric resistivity), real time electric resistivity to 5.3.calculate Instrumentation changes in and unsaturated Remote Sensing water Techniques contents, TLS and aerial laser scanning (ALS); among others [44–47]. Some of the in-place instrumentation at the Ripley Landslide are Only a subset of these landslides have been instrumented, most with slope inclinome- shown in Figure 9. These instruments have made it possible to arrive at the interpretation ters that are manually read, annually. The Ripley landslide, however, has become a ﬁeld of the landslide kinematics as described above. laboratory for investigation and monitoring. Monitoring instrumentation at this landslide Large-scale Satellite Interferometric Synthetic Aperture Radar (InSAR) was used in include grouted-in vibrating wire piezometers, inclinometer casings, Shape Accel Arrays this section of the Thomson River Valley to understand the state of activity of slopes that (SAA), Differential GPS systems, and a wave-guide system that uses waves emitted as the are not being monitored at frequent as the Ripley Landslide, and potentially detect other landslide deforms and interprets them as relative displacement; among others. Site investi- active slopes not being monitored. The images used were captured by the Canadian gation techniques deployed at the site include borehole logging, geophysical investigations (seismic,satellite RADARSAT-2. electric resistivity), East-facing real time images electric (group resistivity F4N) to were calculate collected changes between in unsaturated 2011 and water2015, and contents, west-facing TLS and images aerial (groups laser scanning U5 and U21) (ALS); were among collected others between [44–47 2013]. Some and of 2015. the in-placeImages were instrumentation acquired every at the24 days, Ripley with Landslide some exceptions are shown when in Figure longer9. gaps These occurred. instru- mentsSAR images have made were it possibleprocessed to arriveusing at Persiste the interpretationnt Scatterer of theInterferometry landslide kinematics (PSI), and as describeddisplacements above. measured are the average displacements for areas up to 15 m × 15 m [48].

Figure 9. Line of Sight (LOS) displacement rates from Satellite InSAR at the active landslides (a,b). Location of some in- Figureplace instrumentation 9. Line of Sight (LOS)at the displacementRipley Landslide rates (GSE from are Satellite corner InSAR reflecto at thers, MSE active are landslides manual survey (a,b). Location rods) (c of). Adapted some in-place from instrumentation[44,48]. at the Ripley Landslide (GSE are corner reﬂectors, MSE are manual survey rods) (c). Adapted from [44,48]. Large-scale Satellite Interferometric Synthetic Aperture Radar (InSAR) was used in 5.4. Results and Discussion this section of the Thomson River Valley to understand the state of activity of slopes that areThe not LOS being displacement monitored rate at frequentmeasured as by the the Ripley image Landslide, group U21 and at active potentially landslides detect is othershown active in Figure slopes 9 [48]. not beingThe average monitored. precisio Then imagesof measurements used were capturedwas 1 mm by considering the Canadian the satelliteStandard RADARSAT-2. Deviation of measurements East-facing images at stable (group areas. F4N) were collected between 2011 and

Remote Sens. 2021, 13, 366 13 of 23

2015, and west-facing images (groups U5 and U21) were collected between 2013 and 2015. Images were acquired every 24 days, with some exceptions when longer gaps occurred. SAR images were processed using Persistent Scatterer Interferometry (PSI), and displacements measured are the average displacements for areas up to 15 m × 15 m [48].

5.4. Results and Discussion The LOS displacement rate measured by the image group U21 at active landslides is shown in Figure9[ 48]. The average precision of measurements was 1 mm considering the Standard Deviation of measurements at stable areas. The work in [48] used information about the LOS orientation of the satellite image groups and assumptions of the likely landslide displacement vectors based on the landslide kinematics prevalent in the area. Their analysis identified six areas of concentrated displacements within the study area, all within or adjacent to the footprints of identified landslides (Figure9). Average LOS displacement rates ranged between 11 mm/year (Barnard) and 39 mm/year (Ripley Landslide). The maximum downslope displacement rate (Projection of the LOS to expected landslide displacement vector) was calculated at 120 mm/year (toe of the Goddard Slide). Maximum downslope displacements were between 73 and 89 mm/year for the other landslides, except for the Barnard Slide, with a maximum downslope displacement of 50 mm/year. Use of InSAR allowed the identification of landslides showing an Active state of activity, including the Ripley Landslide and the identification of an active mass adjacent to the South landslide (South Extension in Figure9a), which had not been previously identified. It also allowed the identification of landslides potentially at a Reactivated state (Goddard Toe, Barnard, North Toe, and Red Hill landslides). These insights provide valuable information for an enhanced understanding of the corridor’s risk level associated with potential landslide events and for improved landslide stability analyses for risk assessment purposes [49,50]. The InSAR analysis also identified that most of the Thompson River Valley landslides exhibit seasonal changes in the rate of displacement (median displacements particularly at the Ripley Landslide, Red Hill toe, Goddard toe, and South landslides). These typically show increased displacements during the late fall and winter months versus relative stability during the spring and summer. This suggests that factors that affect the stability of these landslides are common across the study area, and that these landslides are marginally stable [48].

6. Checkerboard Creek Rock Slope 6.1. Location and Geology The Checkerboard Creek rock slope is located within the Revelstoke Reservoir, near the town of Revelstoke, British Columbia, Canada. Figure 10 shows the location of this landslide and a front view from across the reservoir. The typical cross section in Figure 10 presents a sketch of the materials and structure forming the slope. Massive to weakly foliated granodiorite overlie gneiss and schist of a major regional fault system (Columbia River Fault), which has developed a broad, regional, brittle deformation zone of altered and mechanically deformed rock. Shears zones and joints in the slope are frequent, and dip steeply into and out of the slope at angles between 60◦ and 90◦ from horizontal. The rock mass quality ranges from very strong, fresh, undisturbed and blocky rock to highly weathered and altered, weak and disturbed. The lower quality rock masses contain frequent sheared and crushed zones and are typically found within 60 m of the slope surface [11,51]. Remote Sens. 2021, 13, x FOR PEER REVIEW 14 of 22

characteristics. Other instruments and measurements deeper into the active rock mass have shown steady rates of deformation [11,52,53]. The displacement rate is 0.5 mm/year at the boundaries of the actively deforming mass and up to 13 mm/year within the most active area. Deformation rates are greatest at the surface of the slope and decrease progressively, with depths to approximately 50 to 60 m below surface, where no Remote Sens. 2021, 13, 366 deformations are detected. This depth coincides with the weathered and sheared volume14 of 23 of rock described above.

FigureFigure 10. 10.Location Location of of the the Checkerboard Checkerboard Creek Creek rock rock slope slope (a) ( anda) and front front view view (b). (b Typical). Typical cross cross section ofsection the slope of the (c). slope Adapted (c). Adapted from [11 ].from [11].

6.2.6.3. LandslideInstrumentation Extents and and Remote Deformation Sensing Patterns Techniques TheThe CheckerboardCheckerboard Creek Creek rock rock slope isslope 260 mhas high, an from extensive Highway array 23 to theof middlein-place reachinstrumentation, of the Checkerboard mostly installed Creek (Figurebetween 10 1984). The and width 2000. of The the instrumentation slope is 600 m andincludes the ◦ ◦ ◦ overallsurface slope survey angle monuments is 30 , steeper (read atannually), the toe (45 manual) and slope ﬂatter inclinometers in the upper (read area (25 annually))[11]. ◦ Theand active in-place zone probes of deformation (read every has 6 h), an nested average standpipe slope angle piezometers, of 45 , and multipoint deformations borehole have beenextensometers detected down (read toevery 50 to 4 60h), msurface deep. cable The volume extensometers of this active(read annually), zone is estimated strain meters, to be 3 betweenborehole 2 and and surface 3 million thermistors, m . and a weather station that records daily temperature and Slopeprecipitation deformations values are[11,52]. concentrated A global navigation at the central satellite part ofsystem the slope, (GNSS) near network the crest has ofalso the been slope installed cut required at the toslope, accommodate which comprises the highway. of four Displacementssensors. Airborne measured laser scanning by instruments(ALS) was near also thecarried surface out havein September revealed 2014 an annual to acquire cycle a high-resolution dominated by an digital active elevation period frommodel. early October to April/May (early autumn throughout late winter), followed by a relativelyAn quietIBIS-L period GB-InSAR between (ground-based, May and September interferometric, (spring and synthetic summer). aperture These cyclesradar) havemanufactured been interpreted by IDS asGeoradar temperature-dependent, was installed in whereOctober expansion 2016 across and contractionthe Revelstoke of theReservoir rock blocks from nearthe Checkerboard the surface cause Creek changes rock slope. in the The normal system stresses has a at radar the contact head that of jointacquires surfaces, data andas it therefore travels on changing a 2-m-long the rail available to obtain shear a two-dimensional strengths and deformation image from char- the acteristics. Other instruments and measurements deeper into the active rock mass have area of interest (synthetic aperture). The system can target areas at distances of up to 4 shown steady rates of deformation [11,52,53]. The displacement rate is 0.5 mm/year at the boundaries of the actively deforming mass and up to 13 mm/year within the most active area. Deformation rates are greatest at the surface of the slope and decrease progressively, with depths to approximately 50 to 60 m below surface, where no deformations are detected. This depth coincides with the weathered and sheared volume of rock described above. Remote Sens. 2021, 13, 366 15 of 23

6.3. Instrumentation and Remote Sensing Techniques The Checkerboard Creek rock slope has an extensive array of in-place instrumentation, mostly installed between 1984 and 2000. The instrumentation includes surface survey monuments (read annually), manual slope inclinometers (read annually) and in-place probes (read every 6 h), nested standpipe piezometers, multipoint borehole extensometers (read every 4 h), surface cable extensometers (read annually), strain meters, borehole and surface thermistors, and a weather station that records daily temperature and precipitation values [11,52]. A global navigation satellite system (GNSS) network has also been installed at the slope, which comprises of four sensors. Airborne laser scanning (ALS) was also carried out in September 2014 to acquire a high-resolution digital elevation model. An IBIS-L GB-InSAR (ground-based, interferometric, synthetic aperture radar) manu- factured by IDS Georadar was installed in October 2016 across the Revelstoke Reservoir from the Checkerboard Creek rock slope. The system has a radar head that acquires data as it travels on a 2-m-long rail to obtain a two-dimensional image from the area of interest (synthetic aperture). The system can target areas at distances of up to 4 km, covering a horizontal angle of up to 80◦, and vertical angle of up to 40◦. The resolution of the image therefore depends on the distance between the slope and the radar. Challenges for system deployment include the inaccessibility to the electric power grid, which requires a self-sustained power supply. This is compounded with the climate of the site, which includes cold and dark winters, which decrease the solar power potential in the area and decreases the storage of the battery system. Given the very slow deformation rates of the Checkerboard Creek rock slope, and the challenges of power supply at the site, GB-InSAR monitoring was set at to D-InSAR mode (discontinuous, short duration monitoring campaigns), with monitoring campaigns of 1 to 2 weeks long [53]. Within each monitoring campaign, images were acquired at an average frequency of 6 images per hour. Due to the dense vegetation cover, GB-InSAR images were restricted to the exposed rock face adjacent to the highway.

6.4. Results and Discussion The map of displacements in the LOS between the GB-InSAR and the Checkerboard Creek rock slope is shown in Figure 11. The extensive in-place instrumentation at the site allowed interpretation of temporal trends and insights into the spatial deformation patters; however, the GB-InSAR results allowed for an enhanced definition of the active portions of the exposed rock face. This active area detected by the GB-InSAR is bounded by the crest of the exposed rock face and the dark dashed line in Figure 11a. The average deformations in this area are consistent with those measured by in-place instruments. The threshold values used for identifying the most active area was 2 mm/year [52,53]. It was also identified that there is a transition between faster deformation rates near the crest of the expose rock face and its toe. This is consistent with the interpreted deformation kinematics in which the rock blocks within a heavily disaggregated rock mass are showing toppling like behavior and relative displacements between one another [53]. A critical insight from the GB-InSAR monitoring campaign was the identification of a potentially unstable block that appears to be sliding on a ledge defined by a persistent discontinuity that dips 15◦ to 20◦ out of slope (Figure 11). This newly identified potentially unstable block is within an area of the exposed rock face that was previously interpreted as less active. This block was most active between September 2016 and September 2017; however, it did not show increased activity when compared to the surrounding area after that period. This area of the exposed rock face was identified for continuous assessment as part of the slope’s risk management system [53]. Remote Sens. 2021, 13, x FOR PEER REVIEW 15 of 22

km, covering a horizontal angle of up to 80°, and vertical angle of up to 40°. The resolution of the image therefore depends on the distance between the slope and the radar. Challenges for system deployment include the inaccessibility to the electric power grid, which requires a self-sustained power supply. This is compounded with the climate of the site, which includes cold and dark winters, which decrease the solar power potential in the area and decreases the storage of the battery system. Given the very slow deformation rates of the Checkerboard Creek rock slope, and the challenges of power supply at the site, GB-InSAR monitoring was set at to D-InSAR mode (discontinuous, short duration monitoring campaigns), with monitoring campaigns of 1 to 2 weeks long [53]. Within each monitoring campaign, images were acquired at an average frequency of 6 images per hour. Due to the dense vegetation cover, GB-InSAR images were restricted to the exposed rock face adjacent to the highway.

Figure 11. GB-InSAR displacement rate rate in in the LOS at the Checkerboa Checkerboardrd Creek rock slope; crest of exposed rock face is shown with solid dark line, boundary of active deformation is shown in dark dashed line ( a). View of the radar system looking towards the slope (b). D-InSAR D-InSAR average average LOS LOS displacement displacement results results for for the the active active crest crest of the exposed rock face compared againstagainst aa surveysurvey point point and and an an extensometer extensometer located located near near the the crest crest of theof the exposed exposed rock rock face face (c). Adapted(c). Adapted from from [53]. [53].

7. TornadoIt was also Mountain identified and that S042 there Site is a transition between faster deformation rates near the7.1. crest Location of andthe Geologyexpose rock face and its toe. This is consistent with the interpreted deformationTornado kinematics Mountain isin northwhich of the the rock town blocks of Sparwood, within a at heavily the boundary disaggregated between rock the provinces of British Columbia and Alberta, Canada. The study site corresponds to a steep rock outcrop on the British Columbia side, which in reality is part of Mount Lyne, adjacent to Tornado Mountain (Figure 12). The outcrop is 50 m high and near vertical, in strong, blocky limestone above a steep (approximately 45◦) talus slope. The slope of the mountain side becomes gradually softer to an average of 22◦ as it approaches a railway alignment. This lower section of the slope is covered by rock fragments, soil, and is sparsely vegetated. This site has witnessed large blocks (up to 1.5 m in equivalent diameter) rolling downslope to the vicinity of the railway tracks [54]. The S042 Site is located adjacent to Highway 742, approximately 5 km southwest of Canmore, Alberta, Canada (Figure 12). This site consists of a near-vertical, 80-m-high rock slope on limestones, dolomitic limestones, dolostones and shale. Rock falls and erosional processes have created a talus slope of 40◦ between the slope and the highway. Rock falls mostly are captured on the talus slope; however; blocks of less than 0.3 m in equivalent diameter are commonly found on the road [55]. Remote Sens. 2021, 13, 366 17 of 23

Remote Sens. 2021, 13, x FOR PEER REVIEW 17 of 22

FigureFigure 12. 12.Location Location of of the the Tornado Tornado Mountain Mountain study study site site and and S042 S042 site site (a ).(a View). View of of rock rock fall fall source source at theat Tornadothe Tornado Mountain Mountain site (siteb). View(b). View of the of rock the rock slope slope at S042 at S042 Site (Sitec). Adapted (c). Adapted from from [55,56 [55,56].].

7.3.Both Results sites and have Discussion been the matter of rock fall hazard investigations as part of assess- mentsGround-based, into the requirements aerial and and UAV dimensioning photogrammetry of rock fallhave protection become standard structures, practice for the for railwaydeveloping tracks high-resolution and highway surface 742 at themodels Tornado of rock Mountain slopes for and structural Site S042, mapping, respectively. both in Ascivil part and of thesemining investigations, projects. This fragmented information rock is fallthen volume-frequency used as input to relationshipsestimate rock were mass calculatedcharacteristics on the (e.g., basis expected of rock blockbehavior surveys and st andrength) photogrammetric and to develop techniques geotechnical for virtualstability structuraland hydrogeologic mapping (fragmented slope models. rock The falls analyses refer toat fallenTornado block Mountain volumes and after S042 detachment Site are an andexample disaggregation of how this along information their trajectory). can also be used to calculate rock fall hazards in terms of volume-frequency curves. The method and results at Tornado Mountain and the S042 7.2. Photogrammetry and Structural Mapping Site are detailed in [54] and [56]. A similar approach is described in [57] and tested for a rockGround-based fall site in Italy. photogrammetry at the study areas were performed with digital single- lens reflexThe (DSLR) results cameras. of the analyses The sensor at sizeTornado used wasMoun 35.9tain mm are by shown 24 mm, inwith Figure 36.3 million13. This pixels figure. Theshows lens usedthe results at the Tornado of some Mountain steps of site the had analysis, a 200 mm including fixed focal virtual length andmapping the lens of useddiscontinuities for Site S042 on had the a photogrammetric 150 mm fixed focal model length. (Figure The distance 13a,b), betweenas well as the one camera iteration and of thethe slope DFN were developed between 500based m andon 600the m virtual at the Tornadomapping Mountain (Figure site,13c) and and between the resulting 200 m andfragmented 600 m at Siterock S042. fall volume-frequency The setup described curve rendered (Figure ground 13d). Similar pixel sizes results between were obtained 0.7 cm andfor 2the cm S042 [56]. site (Figure 14). This figure shows the resulting discontinuity sets from virtual mappingThe photographs in the photogrammetric at both sites model were processed(Figure 14a), in ADAMone iteration Technology’s of the DFN photogram- developed metry software suite (https://www.adamtech.com.au), which builds a digital surface based on the virtual mapping (Figure 14b), and the resulting fragmented rock fall volume- model using photogrammetric principles and allows automatic and manual detection and frequency curve (Figure 14c). Unfortunately, a view of the structures being mapped mapping of rock structures. Orientations were mapped manually within the software for during the virtual mapping process was not available. both sites. For the two cases analyzed, the rock fall volume-frequency relationships obtained The dip, dip direction and spacing of the sets of discontinuities were fitted to proba- with the DFN’s were consistent with those from surveyed blocks. These results show that bility distributions and used to develop multiple iterations of discrete fracture networks the approach could be used to estimate the fragmented rock fall volume–frequency (DFN) of persistent discontinuities for both sites using Wolfram’s Mathematica software relationships in the absence of rock fall records or block surveys along rock fall package (https://www.wolfram.com). These DFNs were used to calculate the fragmented trajectories, as part of rock fall hazard and risk analyses. The approach presented in [54] rock fall volume-frequency distributions through numerical integration [56]. and [56] was based on the hypothesis that fragmentation in strong rocks (particularly 7.3.limestones Results and characterized Discussion by Unconfined Compressive Strengths over 50 MPa) will tend to occurGround-based, through weak aerial planes andUAV and by photogrammetry extension of non-persistent have become discontinuities standard practice as the for in developingsitu blocks high-resolution detach, fall, bounce, surface and models roll. In of this rock regard slopes the for method structural should mapping, not be both applied in civilto other, and mining weaker projects. rocks, without This information validation. is then used as input to estimate rock mass

Remote Sens. 2021, 13, 366 18 of 23

characteristics (e.g., expected behavior and strength) and to develop geotechnical stability and hydrogeologic slope models. The analyses at Tornado Mountain and S042 Site are an example of how this information can also be used to calculate rock fall hazards in terms of volume-frequency curves. The method and results at Tornado Mountain and the S042 Site are detailed in [54,56]. A similar approach is described in [57] and tested for a rock fall site in Italy. The results of the analyses at Tornado Mountain are shown in Figure 13. This ﬁgure shows the results of some steps of the analysis, including virtual mapping of discontinuities on the photogrammetric model (Figure 13a,b), as well as one iteration of the DFN developed based on the virtual mapping (Figure 13c) and the resulting fragmented rock fall volume-frequency curve (Figure 13d). Similar results were obtained for the S042 site (Figure 14). This ﬁgure shows the resulting discontinuity sets from virtual mapping in the photogrammetric model (Figure 14a), one iteration of the DFN developed based on the virtual mapping (Figure 14b), and the resulting fragmented rock fall volume-frequency curve (Figure 14c). Unfortunately, a view of the structures being mapped during the virtual mapping process was not available. Remote Sens. 2021, 13, x FOR PEER REVIEW 18 of 22

FigureFigure 13.13. VirtualVirtual structural structural mapping mapping (a ()a and) and resulting resulting discontinuity discontinuity sets sets (b) for (b) the for Tornado the Tornado MountainMountain study study site. site. One One iteration iteration of of the th DFNe DFN for for the the Tornado Tornado Mountain Mountain site site (c). ( fragmentedc). fragmented rock rock fall fall volume cumulative distribution based on the DFN realizations compared to block volumes volume cumulative distribution based on the DFN realizations compared to block volumes surveyed surveyed along the rock fall trajectories, DFN volumes are truncated to the minimum range of along the rock fall trajectories, DFN volumes are truncated to the minimum range of surveyed surveyed block volumes for accurate comparison (lack of survey of small fraction of blocks) (d). blockAdapted volumes from for[55,56]. accurate comparison (lack of survey of small fraction of blocks) (d). Adapted from [55,56].

For the two cases analyzed, the rock fall volume-frequency relationships obtained with the DFN’s were consistent with those from surveyed blocks. These results show that the approach could be used to estimate the fragmented rock fall volume–frequency relationships in the absence of rock fall records or block surveys along rock fall trajectories, as part of rock fall hazard and risk analyses. The approach presented in [54,56] was based on the hypothesis that fragmentation in strong rocks (particularly limestones characterized by Unconﬁned Compressive Strengths over 50 MPa) will tend to occur through weak planes and by extension of non-persistent discontinuities as the in situ blocks detach, fall,

Figure 14. Discontinuity sets as a result of the virtual structural mapping (a) for the S042 Site. One iteration of the DFN for the S042 Site (b). Fragmented rock fall volume cumulative distribution based on the DFN realizations compared to block volumes surveyed at the toe of the talus slope, DFN volumes are truncated to the minimum range of surveyed block volumes for accurate comparison (lack of survey of small fraction of blocks) (c). Adapted from [56].

Remote Sens. 2021, 13, x FOR PEER REVIEW 18 of 22

Figure 13. Virtual structural mapping (a) and resulting discontinuity sets (b) for the Tornado Remote Sens. 2021, 13, 366 Mountain study site. One iteration of the DFN for the Tornado Mountain site (c). fragmented19 rock of 23 fall volume cumulative distribution based on the DFN realizations compared to block volumes surveyed along the rock fall trajectories, DFN volumes are truncated to the minimum range of surveyed block volumes for accurate comparison (lack of survey of small fraction of blocks) (d). Adaptedbounce, from and roll.[55,56]. In this regard the method should not be applied to other, weaker rocks, without validation.

Figure 14. Discontinuity sets as a resultresult ofof thethe virtualvirtual structuralstructural mapping mapping ( a(a)) for for the the S042 S042 Site. Site. One One it- iterationeration of of the the DFN DFN for for the the S042 S042 Site Site (b ().b). Fragmented Fragmented rock rock fall fall volume volume cumulative cumulative distribution distribution based basedon the on DFN the realizations DFN realizations compared compared to block to volumesblock volumes surveyed surveyed at the toe at the of the toetalus of the slope, talusDFN slope, vol- DFNumes volumes are truncated are truncated to the minimum to the minimum range of surveyedrange of surveyed block volumes block forvolumes accurate for comparisonaccurate (lack comparison (lack of survey of small fraction of blocks) (c). Adapted from [56]. of survey of small fraction of blocks) (c). Adapted from [56].

8. General Discussion and Conclusions Monitoring plays a fundamental role in landslide risk management. Not only is robust monitoring the cornerstone of effective early warning, but also fundamental for increased understanding of the landslide mechanisms, kinematics and triggers. There is extensive experience in landslide site characterization and monitoring with the use of in situ testing, instrumentation, and remote sensing techniques; and technology keeps evolving such that instruments become more accurate and reliable. In this regard, remote sensing technologies have continued to develop to the point where they are becoming routine practice as part of landslide hazard and risk assessments and management. These include ground-based and airborne (from manned and unmanned aircrafts) photogrammetry, laser scanning, and satellite and ground-based radar interferometry [21–28]. This manuscript takes advan- tage of case studies from the authors’ experience to illustrate the role of remote sensing approaches for implementing the systematic landslide risk management framework that has been widely adopted by industry [8–10,58]. Transportation corridors in Western Canada have to traverse the Canadian Cordillera and river valleys in the prairies, in which landslide hazards are common. The importance of these transportation corridors for the Canadian economy, and their vast extent, require cost effective techniques for landslide investigation and monitoring. Remote sensing techniques have been found to play a key role in achieving this. This paper synthesized six case studies in Western Canada from the authors’ experience in which remote sensing was used in combination with other site investigation and monitoring technologies to gain understanding of the slope deformation mechanisms and develop early warning systems. These case studies cover different landslide types, including rock falls, compound soil and rock landslides, rock mass landslides, translational soil landslides and earth ﬂows. Remote sensing techniques covered in these case studies include ground-based and satellite Remote Sens. 2021, 13, 366 20 of 23

InSAR, ground-based and UAV photogrammetry, as well as photogrammetry on historical photos, and ground-based laser scanning. This illustrates the versatility of remote sensing techniques for landslide investigation showing diverse kinematics and deformation mechanisms, as well as the diversity of available practical remote sensing techniques. The case studies presented here demonstrate the deployment of remote sensing techniques for landslide investigation and monitoring, and how the most gain is obtained when complementing these techniques with information from other, more traditional in-place instrumentation (e.g., slope inclinometers, GPS units, survey prisms). The insights gained in these six case studies, including better definition of landslide deformation trends, both temporally and spatially, improvements in the understanding of landslide mechanisms, insights into rock fall volume-frequency distributions, and identification of the state of activity of large landside at a regional extent, among others. The insights illustrated in this manuscript are based on projects in civil engineering works; however, these are consistent with the reliance on remote sensing techniques by the mining industry to manage potential open pit slope failures. In this regard, remote sensing techniques have already been identified as essential for the successful implementation of open pit slope risk management framework [59]. It is the intent of the authors that these case studies will provide useful examples of the use of remote sensing techniques from a practical perspective and in a variety of contexts. It is noted that the work presented here reflects the methods that were considered most suited for the case studies discussed, which may vary in other contexts and for other jurisdictions. This work is not an exhaustive literature review of remote sensing techniques for landslide investigation; however, they present valuable examples for how the information captured from remote sensing techniques enhance the implementation of the landslide risk management framework.

Author Contributions: Both authors have contributed substantially to the manuscript and the work presented here. R.M. developed the conceptualization of the manuscript and selection of case studies. Both authors (R.M., M.T.H.) have provided the analysis presented in this paper, draft preparation, and visualization, as presented here. All authors have read and agreed to the published version of the manuscript. Funding: The research work referenced in this review article was conducted through the (Canadian) Railway Ground Hazard Research Program, which is funded by the National Sciences and Engineer- ing Research Council of Canada (NSERC ALLRP 549684-19), Canadian Pacific Railway, Canadian Na- tional Railway, and Transport Canada; and through a collaboration with Klohn Crippen Berger Ldt. and Alberta Transportation, funded by NSERC (Engage Gant # EGP 521877-17; KCB CRD # CRDPJ 543429-19) and KCB Engineering Inc. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Acknowledgments: The authors wish to acknowledge Canadian Pacific Railway, Canadian National Railway, Klohn Crippen Berger Ltd., Alberta Transportation and BC Hydro for their contributions to the work presented in this paper and for facilitating access to the sites, historic information and their invaluable experience. In particular, we thank the authors and co-authors of the references cited, which make the basis of the discussions and insights presented here. Conflicts of Interest: The authors declare no conflict of interest.

References 1. Evans, S.G.; Hungr, O. The assessment of rock fall hazard at the base of talus slopes. Can. Geotech. J. 1993, 30, 620–636. [CrossRef] 2. Bunce, C.M.; Cruden, D.M.; Morgenstern, N.R. Assessment of the hazard from rock fall on a highway. Can. Geotech. J. 1997, 34, 344–356. [CrossRef] 3. Hungr, O.; Evans, S.G.; Hazzard, J. Magnitude and frequency of rock falls and rock slides along the main transportation corridors of southwestern British Columbia. Can. Geotech. J. 1999, 36, 224–238. [CrossRef] 4. Clague, J.J.; Evans, S.G. Formation and Failure of Natural Dams in the Canadian Cordillera; Geological Survey of Canada Bulletin: Ottawa, ON, Canada, 1994; Volume 464, 35 p. Remote Sens. 2021, 13, 366 21 of 23

5. Mansour, M.F.; Morgenstern, N.R.; Martin, C.D. Expected damage from displacement of slow-moving slides. Landslides 2011, 8, 117–131. [CrossRef] 6. Macciotta, R.; Cruden, D.; Martin, C.D.; Morgenstern, N.R.; Petrov, M. Spatial and temporal aspects of slope hazards along a railroad corridor in the Canadian Cordillera. In Slope Stability 2013, Proceedings of the International Symposium on Slope Stability in Open Pit Mining and Civil Engineering, Brisbane, Australia, 25–27 September 2013; Dight, P.M., Ed.; Australian Centre for Geomechanics: Perth, WA, Australia, 2013; pp. 1171–1186. 7. Fell, R.; Hartford, D. Landslide risk management. In Landslide Risk Assessment, Proceedings of the International Workshop on Landslide Risk Assessment, Honolulu, Hawaii, 19–21 February 1997; Cruden, D., Fell, R., Eds.; Balkema: Rotterdam, The Netherlands, 1997; pp. 51–108. 8. IUGS Working Group on Landslides; Committee on Risk Assessment. Quantitative risk assessment for slopes and landslides— The state of the art. In Landslide Risk Assessment, Proceedings of the International Workshop on Landslide Risk Assessment, Honolulu, Hawaii, 19–21 February 1997; Cruden, D., Fell, R., Eds.; Balkema: Rotterdam, The Netherlands, 1997; pp. 3–12. 9. Ho, K.K.S.; Leroi, E.; Roberds, W.J. Quantitative risk assessment application, myths and future direction. In GeoEng2000, Proceedings of the International Conference on Geotechnical and Geological Engineering, Melbourne, Australia, 19–24 November 2000; Technomic: Chicago, IL, USA, 2000; pp. 269–312. 10. Fell, R.; Ho, K.K.S.; Lacasse, S.; Leroi, E. A framework for landslide risk assessment and management. In Landslide Risk Management, Proceedings of the International Conference on Landslide Risk Management, Vancouver, BC, Canada, 31 May–3 June 2005; Hungr, O., Fell, R., Couture, R., Eberhardt, E., Eds.; CRC Press: Boca Raton, FL, USA, 2005; pp. 3–25. 11. Macciotta, R.; Martin, C.D.; Morgenstern, N.R.; Cruden, D.M. Development and application of a quantitative risk assessment to a very slow moving rock slope and potential sudden acceleration. Landslides 2016, 13, 765–785. [CrossRef] 12. Morgenstern, N.R. Toward landslide risk assessment in practice. In Landslide Risk Assessment, Proceedings of the International Workshop on Landslide Risk Assessment, Honolulu, Hawaii, 19–21 February 1997; Balkema: Rotterdam, The Netherlands, 1997; pp. 15–23. 13. Mostyn, G.; Sullivan, T. Quantitative risk assessment of the Thredbo landslide. Aust. Geomech. 2002, 37, 169–181. 14. Bonnard, C.; Forlati, F.; Scavia, C. (Eds.) Identification and Mitigation of Large Landslide Risks in Europe—Advances in Risk Assessment; Taylor and Francis Group: London, UK, 2004. 15. Lacasse, S.; Eidsvik, U.; Nadim, F.; Høeg, K.; Blikra, L.H. Event tree analysis of Aknes Rockslide hazard. In Proceedings of the 4th Canadian Conference on Geohazards: From Causes to Management, Canadian Geotechnical Society, Quebec City, QC, Canada, 20–24 May 2008; pp. 551–557. 16. Zhao, C.; Lu, Z. Remote sensing of landslides—A review. Remote Sens. 2018, 10, 279. [CrossRef] 17. Bianchini, S.; Herrera, G.; Mateos, R.M.; Notti, D.; Garcia, I.; Mora, O.; Moretti, S. Landslide activity maps generation by means of persistent scatterer interferometry. Remote Sens. 2013, 5, 6198–6222. [CrossRef] 18. Atzeni, C.; Barla, M.; Pieraccini, M.; Antolini, F. Early warning monitoring of natural and engineered slopes with ground-based synthetic-aperture radar. Rock Mech. Rock Eng. 2014, 48, 235–246. [CrossRef] 19. Tofani, V.; del Ventisette, C.; Moretti, S.; Casagli, N. Integration of remote sensing techniques for intensity zonation within a landslide area: A case study in the Northern Apennines, Italy. Remote Sens. 2014, 6, 907–924. [CrossRef] 20. Zhang, Y.; Meng, X.; Chen, G.; Qiao, L.; Zeng, R.; Chang, J. Detection of geohazards in the Bailong River basin using synthetic aperture radar interferometry. Landslides 2016, 13, 1273–1284. [CrossRef] 21. Bardi, F.; Raspini, F.; Frodella, W.; Lombardi, L.; Nocentini, M.; Gigli, G.; Morelli, S.; Corsini, A.; Casagli, N. Monitoring the rapid-moving reactivation of earth flows by means of GB-InSAR: The April 2013 Capriglio landslide (Northern Apennines, Italy). Remote Sens. 2017, 9, 165. [CrossRef] 22. Kromer, R.; Lato, M.; Hutchinson, D.J.; Gauthier, D.; Edwards, T. Managing rockfall risk through baseline monitoring of precursors using a terrestrial laser scanner. Can. Geotech. J. 2017, 54, 953–967. [CrossRef] 23. Lombardi, L.; Nocentini, M.; Frodella, W.; Nolesini, T.; Bardi, F.; Intrieri, E.; Carlà, T.; Solari, L.; Dotta, G.; Ferrigno, F.; et al. The Calatabiano landslide (southern Italy): Preliminary GB-InSAR monitoring data and remote 3D mapping. Landslides 2017, 14, 1–12. [CrossRef] 24. Novellino, A.; Cigna, F.; Sowter, A.; Ramondini, M.; Calcaterra, D. Exploitation of the intermittent SBAS (ISBAS) algorithm with COSMO-SkyMed data for landslide inventory mapping in north-western Sicily, Italy. Geomorphology 2017, 280, 153–166. [CrossRef] 25. Riquelme, A.; Cano, M.; Tomás, R.; Abellán, A. Identification of rock slope discontinuity sets from laser scanner and photogrammetric point clouds: A comparative analysis. Procedia Eng. 2017, 19, 838–845. [CrossRef] 26. Cucchiaro, S.; Cavalli, M.; Vericat, D.; Crema, S.; Llena, M.; Beinat, A.; Marchi, L.; Cazorzi, F. Monitoring topographic changes through 4D-structure-from-motion photogrammetry: Application to a debris-flow channel. Environ. Earth Sci. 2018, 77, 1–21. [CrossRef] 27. Gomez, F.; Rosenblad, B.; Loehr, E.; Smoot, J.; Lowry, B. Long-term monitoring of a slow moving landslide before and after remediation using ground-based radar interferometry. In Proceedings of the 2019 ASCE Geo-Congress GSP 311, Philadelphia, PA, USA, 24–27 March 2019. Remote Sens. 2021, 13, 366 22 of 23

28. Farina, P.; Catani, F.; Rosi, A.; Setiawan, I.; Junaidi, A.; Afrizal, K.; Wijayanto, A. Development of an early warning system for shallow landslide hazard in the Grasberg area, Indonesia. In Proceedings of the 2020 International Symposium on Slope Stability in Open Pit Mining and Civil Engineering, Perth, WA, Australia, 12–14 May 2020; Dight, P.M., Ed.; Australian Centre for Geomechanics: Perth, WA, Australia, 2020; pp. 1425–1438. 29. Deane, E.; Macciotta, R.; Hendry, M.; Gräpel, C.; Skirrow, R. The use and limitations of modern technologies for slow vegetated landslide monitoring—Chin Coulee landslide. In Proceedings of the 72nd Canadian Geotechnical Conference, Geo St. John’s 2019, St. John’s, NL, Canada, 29 September–2 October 2019. 30. Deane, E.; Macciotta, R.; Hendry, M.; Gräpel, C.; Skirrow, R. Leveraging historical aerial photographs and digital photogrammetry techniques for landslide investigation—A practical perspective. Landslides 2020, 17, 1989–1996. [CrossRef] 31. Deane, E.; Macciotta, R.; Hendry, M.; Gräpel, C.; Skirrow, R. Historical aerial photographs and digital photogrammetry techniques to investigate the development and evolution of the Chin Coulee landslide in Alberta. In GeoVirtual 2020, Proceedings of the Canadian Geotechnical Conference, Calgary, AB, Canada, 14–16 September 2020; Klohn Crippen Berger: Calgary, AB, Canada, 2020. 32. Teledyne Optech. ILRIS-LR Specifications Sheet. Available online: http://www.ticgroup.com.tw/menu/products/sur/products/ 3D_Laser/ILRIS%20LR.pdf (accessed on 7 December 2020). 33. Lague, D.; Nicolas, B.; Jérôme, L. Accurate 3D comparison of complex topography with terrestrial laser scanner: Application to the Rangitikei Canyon (N-Z). ISPRS J. Photogramm. Remote Sens. 2013, 82, 10–26. [CrossRef] 34. Roustaei, M.; Macciotta, R.; Hendry, M.T.; Rodriguez, J.; Gräpel, C.; Skirrow, R. Characterization of a rock slope showing three weather-dominated failure modes. In Slope Stability 2020, Proceedings of the International Symposium on Slope Stability in Open Pit Mining and Civil Engineering, Perth, WA, Australia, 12–14 May 2020; Dight, P.M., Ed.; Australian Centre for Geomechanics: Perth, WA, Australia, 2020; pp. 427–438. 35. Rodriguez, J.; Macciotta, R.; Hendry, M.T.; Roustaei, M.; Gräpel, C.; Skirrow, R. UAVs for monitoring, investigation, and mitigation design of a rock slope with multiple failure mechanisms—A case study. Landslides 2020, 17, 2027–2040. [CrossRef] 36. Jenson, S.K.; Domingue, J.O. Extracting topographic structure from digital elevation data for geographic information system analysis. Photogramm. Eng. Remote Sens. 1988, 54, 1593–1600. 37. Macciotta, R.; Martin, C.D.; Cruden, D.M. Probabilistic estimation of rockfall height and kinetic energy based on a three- dimensional trajectory model and Monte Carlo simulation. Landslides 2015, 12, 757–772. [CrossRef] 38. Macciotta, R.; Martin, C.D. Preliminary approach for prioritizing resource allocation for rock fall hazard investigations based on susceptibility mapping and efficient three-dimensional trajectory modelling. Bull. Eng. Geol. Environ. 2019, 78, 2803–2815. [CrossRef] 39. Carlà, T.; Macciotta, R.; Hendry, M.; Martin, C.D.; Edwards, T.; Evans, T.; Farina, P.; Intrieri, E.; Casagli, N. Displacement of a landslide retaining wall and application of an enhanced failure forecasting approach. Landslides 2018, 15, 489–505. [CrossRef] [PubMed] 40. Macciotta, R.; Rodriguez, J.; Hendry, M.T.; Martin, C.D.; Edwards, T.; Evans, T. The 10-mile slide North of Lillooet, British Columbia—History, characteristics and monitoring. In Proceedings of the 3rd North American Symposium on Landslides, Roanoke, VA, USA, 4–8 June 2017; Association of Environmental & Engineering Geologists (AEG): Brunswick, OH, USA, 2017; pp. 937–948. 41. Rodriguez, J.; Hendry, M.T.; Macciotta, R.; Evans, T. Cost-effective landslide monitoring GPS system: Characteristics, implementation and results. In Geohazards 7, Proceedings of the 7th Canadian Geohazards Conference, Canmore, AB, Canada, 3–6 June 2018. Available online: https://geohazardassociation.org/event/geohazards-7/ (accessed on 7 December 2020). 42. Hendry, M.T.; Macciotta, R.; Martin, C.D.; Reich, B. Effect of Thompson River elevation on velocity and instability of Ripley Slide. Can. Geotech. J. 2015, 52, 257–267. [CrossRef] 43. Clague, J.J.; Evans, S.G. Geologic framework of large historic landslides in Thompson River Valley, British Columbia. J. Environ. Eng. Geosci. 2003, 9, 201–212. [CrossRef] 44. Macciotta, R.; Hendry, M.; Martin, C.D. Developing and early warning system for a very slow landslide based on displacement monitoring. Nat. Hazards 2016, 81, 887–907. [CrossRef] 45. Macciotta, R.; Hendry, M.; Martin, C.D.; Elwood, D.; Lan, H.; Huntley, D.; Bobrowsky, P.; Sladen, W.; Bunce, C.; Choi, E.; et al. Monitoring of the Ripley landslide in the Thompson River Valley, B.C. In Proceedings of the 6th Canadian Geohazards Conference, Kingston, ON, Canada, 15–18 June 2014. 46. Huntley, D.; Bobrowsky, P.; Hendry, M.; Macciotta, R.; Best, M. Multi-technique geophysical investigation of a very slow-moving landslide near Ashcroft, British Columbia, Canada. J. Environ. Eng. Geophys. 2019, 24, 87–110. [CrossRef] 47. Huntley, D.; Bobrowsky, P.; Hendry, M.; Macciotta, R.; Elwood, D.; Sattler, K.; Best, M.; Chambers, J.; Meldrum, P. Application of multi-dimensional electrical resistivity tomography datasets to investigate a very slow-moving landslide near Ashcroft, British Columbia, Canada. Landslides 2019, 16, 1033–1042. [CrossRef] 48. Journault, J.; Macciotta, R.; Hendry, M.T.; Charbonneau, F.; Huntley, D.; Bobrowsky, P. Measuring displacements of the Thompson River valley landslides, south of Ashcroft, BC, Canada, using satellite InSAR. Landslides 2018, 15, 621–636. [CrossRef] 49. De Novellis, V.; Castaldo, R.; Lollino, P.; Manunta, M.; Tizzani, P. Advanced three-dimensional finite element modeling of a slow landslide through the exploitation of DInSAR measurements and in situ surveys. Remote Sens. 2016, 8, 670. [CrossRef] 50. Barra, A.; Solari, L.; Béjar-Pizarro, M.; Monserrat, O.; Bianchini, S.; Herrera, G.; Crosetto, M.; Sarro, R.; González-Alonso, E.; Mateos, R.M.; et al. A methodology to detect and update active deformation areas based on sentinel-1 SAR images. Remote Sens. 2017, 9, 1002. [CrossRef] Remote Sens. 2021, 13, 366 23 of 23

51. Stewart, T.W.G.; Moore, D.P. Displacement behaviour of the Checkerboard Creek rock slope. In Terrain Stability in the Interior of British Columbia; Technical Report 003; BC Ministry of Forests: Dawson Creek, BC, Canada, 2002. 52. Woods, A.; Hendry, M.T.; Macciotta, R.; Stewart, T.; Marsh, J. GB-InSAR monitoring of vegetated and snow-covered slopes in remote mountainous environments. Landslides 2020, 17, 1713–1726. [CrossRef] 53. Woods, A.; Hendry, M.T.; Macciotta, R.; Stewart, T.; Marsh, J. Updated understanding of the deformation characteristics of the Checkerboard Creek rock slope through GB-InSAR monitoring. Eng. Geol. 2021, 281, 105974. [CrossRef] 54. Macciotta, R.; Martin, C.D. Remote structural mapping and discrete fracture networks to calculate rock fall volumes at Tornado Mountain, British Columbia. In Proceedings of the 49th US Rock Mechanics/Geomechanics Symposium, San Francisco, CA, USA, 28 June–1 July 2015. 55. Macciotta, R.; Gräpel, C.; Keegan, T.; Duxbury, J.; Skirrow, R. Quantitative risk assessment of rock slope instabilities that threaten a highway near Canmore, Alberta, Canada: Managing risk calculation uncertainty in practice. Can. Geotech. J. 2020, 57, 337–353. [CrossRef] 56. Macciotta, R.; Gräpel, C.; Skirrow, R. Fragmented rockfall volume distribution from photogrammetry-based structural mapping and discrete fracture networks. Appl. Sci. 2020, 10, 6977. [CrossRef] 57. Francioni, M.; Antonaci, F.; Sciarra, N.; Robiati, C.; Coggan, J.; Stead, D.; Calamita, F. Application of unmanned aerial vehicle data and discrete fracture network models for improved rockfall simulations. Remote Sens. 2020, 12, 2053. [CrossRef] 58. Lacasse, S. Hazard, reliability and risk assessment—Research and practice for increased safety. In Proceedings of the 17th Nordic Geotechnical Meeting, Reykjavik, Iceland, 25–28 May 2016; pp. 17–42. 59. Sharon, R.; Eberhardt, E. Guidelines for Slope Performance Monitoring; CRC Press, Taylor & Francis Group: Leiden, The Netherlands, 2020.